International Journal of Industrial Ergonomics 47 (2015) 1e8

Contents lists available at ScienceDirect

International Journal of Industrial Ergonomics

journal homepage: www.elsevier.com/locate/ergon

Revised NIOSH Lifting Equation May generate spine loads exceeding

recommended limits
Navid Arjmand a, *, Mohammad Amini a, 1, Aboulfazl Shirazi-Adl b, 2, Andre

Plamondon c, 3,
Mohammad Parnianpour a, 4
a
Department of Mechanical Engineering, Sharif University of Technology, Tehran, Iran
b

Division of Applied Mechanics, Department of Mechanical Engineering, Ecole Polytechnique, Montr

eal, Qu

ebec, Canada
c
Institut de recherche Robert Sauv

e en sant

e et en s

ecurit

e du travail, Montr

eal, Qu

ebec, Canada

a r t i c l e i n f o a b s t r a c t

Article history: The 1991 NIOSH Lifting Equation (NLE) is widely used to assess the risk of injury to spine by providing
Received 27 April 2014 estimates of the recommended weight limit (RWL) in hands. The present study uses the predictive
Received in revised form equations developed based on a detailed trunk musculoskeletal biomechanical model to verify whether
5 September 2014
the RWL generates L5-S1 loads within the limits (e.g., 3400 N for compression recommended by NIOSH
Accepted 21 September 2014
Available online 27 February 2015
and 1000 N for shear recommended in some studies). Fifty lifting activities are simulated here to evaluate
the RWL by the NLE and the L5-S1 loads by the predictive equations. In lifting activities involving
moderate to large forward trunk flexion, the estimated RWL generates L5-S1 spine loads exceeding the
Keywords:
NIOSH lifting equation
recommended limits. The NIOSH vertical multiplier is the likely cause of this inadequacy; a revised
Biomechanical model multiplier accounting for the trunk flexion angle is hence needed. The use of a fixed 3400 N compression
Spine loads limit is also questioned.
Multiplier Relevance to industry: Ergonomist and occupational health practitioners are advised to use the NLE along
Recommended weight limit with the predictive equations of the spine loads proposed in this study when evaluating the risk of injury
to the spine in lifting activities particularly those involving moderate to large trunk flexion.
© 2015 Elsevier B.V. All rights reserved.

1. Introduction manage the increasing rate of work-related LBP, the National

Institute for Occupational Safety and Health (NIOSH) has pro-
Epidemiological studies have identified manual material posed the NIOSH Lifting Equation (NLE) (Waters et al., 1993). For a
handling (MMH) and lifting activities as risk factors for low back given manual lifting activity, the NLE estimates the recommended
pain (LBP) (Garg and Moore, 1992; Hoogendoorn et al., 2000; weight limit (RWL) that almost all healthy workers (90% males
Manchikanti, 2000; Van Nieuwenhuyse et al., 2004). Both and 75% females) may handle without an increased risk of LBP.
modeling studies, by predicting increased compression, shear and The lifting index (LI) is subsequently defined as the ratio of the
moment loads on the intervertebral discs (Arjmand and Shirazi- load lifted by the worker in the workplace to the RWL computed
Adl, 2006), and in vivo investigations, by measuring higher by the NLE.
intradiscal pressures (Nachemson, 1981; Wilke et al., 2001), In the NLE, the RWL (in kg) is estimated by modulating a weight
corroborate this association between LBP and MMH. In order to constant (23 kg) by six task-related multipliers varying between
0 and 1; RWL ¼ 23 kg  HM  VM  AM  DM  CM  FM where
HM, VM, AM, DM, CM, and FM are respectively, horizontal, vertical,
* Corresponding author. Sharif University of Technology, Tehran, 11155-9567, asymmetry, distance, coupling, and frequency multipliers that, in
Iran. Tel.: þ98 21 6616 5684; fax: þ98 21 6600 0021. turn, are respectively computed based on the horizontal (H) and
E-mail addresses: [email protected] (N. Arjmand), mohammad.amini89@ vertical (V) positions of the handled load with respect to the inter-
gmail.com (M. Amini), [email protected] (A. Shirazi-Adl), plamondon.
feet midpoint, handled load asymmetry to the body's midesagittal
[email protected] (A. Plamondon), [email protected] (M. Parnianpour).
1
Tel.: þ98 912 299 7824. plane (A), vertical travel distance of the lift (D), hand to load
2
Tel.: þ1 514 340 4711x4129; fax: þ1 514 340 4176. coupling (C), and frequency of lift (F). Three distinct criteria are
3
Tel.: þ1 514 288 1551x279; fax: þ1 514 288 6097. considered when estimating the NLE multipliers (e.g., relationship
4
Tel.: þ98 21 6616 5525; fax: þ98 21 6600 0021.

http://dx.doi.org/10.1016/j.ergon.2014.09.010
0169-8141/© 2015 Elsevier B.V. All rights reserved.
2 N. Arjmand et al. / International Journal of Industrial Ergonomics 47 (2015) 1e8

between HM and H): physiological (limiting maximum energy models for the estimation of the L5-S1 compressive loads (Waters
expenditure to 2.2e4.7 kcal/min), psychophysical (limiting weight et al., 1993) and since biomechanical models have thus far
magnitude to a value acceptable to 75% of female workers), and evolved significantly, it is important to critically re-evaluate the
biomechanical (limiting L5-S1 compressive force to 3400 N) biomechanical accuracy of the NLE in determining the RWL.
(Waters et al., 1993). The NIOSH committee has considered the L5-S1 compressive
The NLE is widely used worldwide by occupational health force as the only critical stress factor when determining the RWL
practitioners to assess the risk of LBP (Waters et al., 1998). A web- while the posterior-anterior shear forces on the lumbar interver-
based survey in Canada indicates that most of the certified ergon- tebral discs during lifting activities likely also act as a risk factor for
omists use the NLE in industry to identify and prevent work-related LBP (Norman et al., 1998). Although biomechanical models gener-
musculoskeletal disorders (Pascual and Naqvi, 2008). The adequacy ally predict much lower shear forces when compared to compres-
of the NLE in controlling biomechanical spine loads during lifting sive forces (Arjmand and Shirazi-Adl, 2006; Arjmand et al., 2006)
activities, however, remains unknown. That is, the RWL estimated the spine strength is however also much lower in shear. While yield
by the NLE for a specific lifting activity may indeed generate L5-S1 and peak thresholds of about 1500 N and 3000 N have respectively
compressive force greater than the recommended limit of 3400 N. been measured in vitro in lumbar cadaver motion segments
As the 1991 NIOSH committee used very simplified biomechanical (Skrzypiec et al., 2012) maximum permissible limit of 1000 N is

Table 1
Fifty lifting tasks in upright standing and flexed postures with load held at different horizontal (H) and vertical (V) positions relative to the inter-feet midpoint (tasks 1e2: load
holding in upright posture close to and away from the chest (see Fig. 1a), tasks 3e6: load holding in upright posture at different vertical but identical horizontal distances (see
Fig. 1b), and tasks 7e50: load holding at different flexed postures and horizontal distances (see Fig. 1c for an example)). Corresponding position of the hand load to the L5-S1
joint (DL5-S1), trunk flexion angle (T), and pelvis flexion angle (P) measured in vivo are also given. Predicted NIOSH horizontal (HM) and vertical (VM) multipliers to estimate the
recommended weight limit (RWL) and the L5-S1 compressive and shear forces with the RWL in hands for each tasks are estimated.

Task V (cm) H (cm) DL5-S1 (cm) T (deg) P (deg) HM VM RWL (kg) Compressive load (N) Shear load (N)

1 130 12 25 0 0 1.000 0.835 19.2 1196 705

2 130 40 55 0.625 0.835 12.0 1858 963
3 90 23 30 0 0 1.000 0.955 22.0 1446 920
4 120 1.000 0.865 19.9 1401 845
5 150 1.000 0.775 17.8 1346 771
6 180 1.000 0.685 15.8 1282 699
7 79.5 25 35.6 10 2.6 1.000 0.987 22.7 3144 1197
8 30 40.6 0.833 18.9 2985 1142
9 40 50.6 0.625 14.2 2835 1092
10 50 60.6 0.500 11.3 2794 1081
11 77.9 25 37.7 20 6.9 1.000 0.991 22.8 3535 1278
12 30 42.7 0.833 19.0 3379 1222
13 40 52.7 0.625 14.2 3233 1171
14 50 62.7 0.500 11.4 3195 1159
15 75.3 25 39.8 30 9.5 1.000 0.999 23.0 3801 1329
16 30 44.8 0.833 19.1 3651 1280
17 40 54.8 0.625 14.4 3509 1238
18 50 64.8 0.500 11.5 3468 1230
19 71.4 25 42.2 40 12.4 1.000 0.989 22.8 4046 1368
20 30 47.2 0.833 19.0 3896 1324
21 40 57.2 0.625 14.2 3750 1285
22 50 67.2 0.500 11.4 3704 1279
23 66.1 25 45.0 50 14.4 1.000 0.973 22.4 4265 1384
24 30 50.0 0.833 18.7 4106 1342
25 40 60.0 0.625 14.0 3947 1306
26 50 70.0 0.500 11.2 3890 1299
27 60.6 25 47.1 60 16.8 1.000 0.957 22.0 4430 1379
28 30 52.1 0.833 18.3 4264 1338
29 40 62.1 0.625 13.8 4091 1301
30 50 72.1 0.500 11.0 4022 1294
31 53.7 25 48.5 70 22.2 1.000 0.936 21.5 4565 1396
32 30 53.5 0.833 17.9 4389 1352
33 40 63.5 0.625 13.5 4203 1311
34 50 73.5 0.500 10.8 4125 1301
35 47.1 25 49.9 80 28.3 1.000 0.916 21.1 4677 1414
36 30 54.9 0.833 17.6 4487 1366
37 40 64.9 0.625 13.2 4282 1320
38 50 74.9 0.500 10.5 4192 1306
39 40.7 25 51.2 90 35 1.000 0.897 20.6 4752 1421
40 30 56.2 0.833 17.2 4545 1369
41 40 66.2 0.625 12.9 4317 1316
42 50 76.2 0.500 10.3 4212 1298
43 32.9 25 50.7 100 41 1.000 0.874 20.1 4671 1356
44 30 55.7 0.833 16.7 4458 1303
45 40 65.7 0.625 12.6 4220 1249
46 50 75.7 0.500 10.0 4106 1230
47 27.7 25 49.4 107.6 46 1.000 0.858 19.7 4549 1284
48 30 54.4 0.833 16.4 4335 1233
49 40 64.4 0.625 12.3 4094 1180
50 50 74.4 0.500 9.9 3976 1160
 N. Arjmand et al. / International Journal of Industrial Ergonomics 47 (2015) 1e8 3

recommended in some studies for spine shear forces in single ex- trunk flexion with a stoop lift, i.e., knee straight and lumbar flexed)
ertions (Gallagher and Marras, 2012; McGill, 1997; McGill et al., at different horizontal distances were studied (Table 1 and Fig. 1).
1998). Therefore, for a given lifting activity even if the RWL pro- Kinematics data from our previous in vivo studies (Arjmand et al.,
posed by the NLE adequately controls the in vivo L5-S1 compressive 2009, 2010) collected on a healthy male subject (52 years,
force, it is not clear whether the shear forces remain bounded 174.5 cm, and 68.4 kg) under these lifting tasks were used to
within equally safe limits. determine the horizontal (H) and vertical (V) position of the
This study, hence, aims to investigate the adequacy of the NLE in handled load with respect to the inter-feet midpoint needed in the
controlling in vivo biomechanical loads of the spine during infre- NLE for the estimation of the RWL. For these tasks, corresponding
quent (FM ¼ 1) sagittally-symmetric (AM ¼ 1) manual handling body posture including sagittal trunk (T) and pelvis (P) rotations
activities with good hand to load coupling (CM ¼ 1) and small with respect to the upright posture as well as horizontal position
vertical travel distance of the lift (DM ¼ 1). For this purpose, a (DL5-S1) of the handled load with respect to the L5-S1 needed for
number of in vivo symmetric lifting activities in upright and flexed the estimation of spinal loads using the predictive equations were
postures are used to evaluate the horizontal (H) and vertical (V) also determined. Kinematics was measured using 12 clusters; one
position of the handled load needed both to compute HM and VM on each foot, thigh, upper arm and forearm in addition to one on
multipliers for the calculation of RWL and to determine body the pelvis, T12, C7, head, and handled load. Four light-emitting
posture required for biomechanical analysis of spine loads. This diodes (LEDs) markers were glued on each cluster (except for the
RWL is subsequently considered as input into our predictive feet with seven LEDs) for a total of 54 LEDs. Positions of the markers
equations of the spine loads (Arjmand et al., 2011) under identical were measured in three dimensions at a sampling rate of 30 Hz
lifting posture measured in vivo. The predictive equations used to using a four-camera Optotrak system (Northern Digital, Waterloo
estimate compressive and shear loads on the L5-S1 intervertebral ON, Canada). The study was limited to the simulation of infrequent
disc during lifting activities are accurate easy-to-use tools devel- lifting tasks performed symmetrically in the sagittal plane, at
oped based on a validated finite element musculoskeletal biome- relatively slow movement speed. Lifting tasks in the upright
chanical model of the spine (Arjmand and Shirazi-Adl, 2006; posture with load held at two different horizontal distances (tasks 1
Arjmand et al., 2007, 2009, 2010). It is hypothesized that under and 2 in Table 1), in the upright posture with load held at four
some RWLs the spinal loads likely exceed the recommended safe different vertical positions (tasks 3 through 6 in Table 1) and in
compression and shear force levels. flexed postures (Tasks 7 to 50 in Table 1) were performed with
respectively 19.8, 10.4, and 0 kg in hands in our previous in vivo
studies (Arjmand et al., 2009, 2010). Weight in hands were
2. Materials and methods comfortably grasped via handles.

2.1. In vivo study

2.2. Recommended weight limit (RWL)
Fifty sagittally-symmetric manual lifting activities while hold-
ing a weight in upright posture at different horizontal and vertical For each lifting task, based on the measured horizontal (H) and
distances as well as in different flexed postures (from 10 to full vertical (V) distance of the hand load with respect to the midpoint

Fig. 1. A schematic presentation of the tasks performed by the subject in the in vivo study to determine position of load and body posture: (a) holding symmetrically a box of 19.8 kg
close to and far away from the chest (b) holding symmetrically a bar of 10.4 kg at four different heights (90, 120, 150, and 180 cm with respect to the ground) while preserving its
horizontal moment arm at 30 cm with respect to the S1, and (c) an example of tasks performed with forward trunk flexion (T ¼ 40 ).
4 N. Arjmand et al. / International Journal of Industrial Ergonomics 47 (2015) 1e8

between the feet (Table 1), the horizontal (HM) and vertical (VM) (Table 1). In practical application, one needs to only measure trunk
multipliers were calculated using the NIOSH proposed relation- rotation while the LP ratio is estimated based on the available
ships between H (in cm) and HM, i.e. HM ¼ 25/H (except for database in the literature (Arjmand et al., 2011; Tafazzol et al.,
H < 25 cm where HM ¼ 1), as well as V (in cm) and VM, i.e. 2014). To determine load distance to shoulder joint (DS) the
VM ¼ 1(0.003  jV75j) (Waters et al., 1993). Other NIOSH following equations developed based on our model posture pre-
multipliers including asymmetry (AM), distance (DM), coupling diction were used for the tasks in upright and flexed postures,
(CM), and frequency (FM) multipliers were taken as their reference respectively (Arjmand et al., 2011):
values (AM ¼ DM ¼ CM ¼ FM ¼ 1).
DS ðUprightÞ ¼ DL5S1 þ 5:271 þ 0:0229  M

2.3. Predictive equations for biomechanical analysis of spine loads

DS ðFlexedÞ ¼ DL5S1 þ 6:0682  0:8205  T þ 1:301  LP
Predictive equations are simple yet accurate user-friendly tools þ 0:0035  T2  0:2551  LP2 þ 0:007  T  LP
that relate outputs (spine compressive and shear loads) of a com-
plex detailed trunk finite element biomechanical model of the
where DL5-S1 is load distance to the L5-S1 joint centre in cm as given
spine to its input variables during static lifting activities (Arjmand
in Table 1. For each of fifty lifting activities the magnitude of T, M
et al., 2011). Sagittal trunk flexion angle (T), lumbopelvic ratio
(equal to the estimated RWL by the NLE), LP, and DL5-S1 as given in
(LP), mass (M) of the handled load and its horizontal distance (DS)
Table 1 were replaced in the appropriate predictive equations and
with respect to the shoulder joint are the model inputs. These
the L5-S1 spine loads were estimated. Subsequently, the maximum
equations are developed to serve ergonomists and occupational
weight in hands (M3400) that results in the L5-S1 compression force
health practitioners in management of low back disorders in esti-
of 3400 N (calculated by trial-and-error using the predictive
mation of spinal loads and design of workplace. Validity of the
equations) was compared to the RWL for each task.
predictive equations has been verified (Arjmand et al., 2011) by
comparing their predictions for spine compressive loads under a
3. Results
number of lifting activities with the measured intradiscal pressure
values under identical tasks (R2 ¼ 0.99 and RMSE ¼ 0.12 MPa). As
The NLE adequately controlled the L5-S1 spinal loads during
for the estimation of the L5-S1 compressive and shear forces during
lifting activities in upright standing posture by yielding RWLs that
lifting tasks in upright and flexed postures four equations were
correspond to compressive loads far smaller than the NIOSH action
used as follows (Arjmand et al., 2011):
limit of 3400 N and shear loads smaller than 1000 N (Table 1: tasks
1 through 6). The NLE is therefore very conservative (Cupright < 2000
Cupright ¼ 407:143 þ 9:088  M  1:457  DS  1:133  M2
N) for these activities as far as the L5-S1 compressive load is con-
þ 0:106  Ds 2 þ 1:657  MDS cerned (Table 1). Also, the NLE adequately controlled the L5-S1
(1) compressive loads to remain lower than 3400 N during almost all
lifting tasks involving small trunk flexion angles (T < 30 ) (tasks 7
through 14 with the exception of task 11 in Table 1). For all lifting
Supright ¼ 93:797  6:621  M þ 2:093  DS þ 0:097  M2 tasks with trunk flexion angles larger than 30 (tasks 15 through
þ 0:009  Ds 2 þ 1:067  MDS 50), however, the proposed RWL of the NLE generated L5-S1
compressive loads up to 40% beyond the NIOSH limit of 3400 N.
(2)
Also, for all lifting tasks in flexed postures the predicted L5-S1 shear
load with the RWL in hands exceeded 1000 N (Table 1). Comparison
Cflexed ¼ 56:435 þ 56:584  T þ 21:6  M  214:667  LP of the maximum weight in hands (M3400) that results in the L5-S1
þ 10:247  DS  0:332  T2 þ 0:238  M2 þ 50:532 compression limit of 3400 N to the RWL for lifting tasks indicated
greater RWL in lifting activities involving trunk flexion angles larger
 ðLPÞ2 þ 0:027  Ds 2 þ 0:764  TM  0:745  TðLPÞ than 30 (i.e., RWL/M3400 > 1 and in some cases close or even larger
 0:053  TDS  3:823  MðLPÞ þ 2:412  MDS than 2) (Table 2).

 0:923  ðLPÞDS
4. Discussion
(3)
Accurate yet easy-to-use predictive equations of the spine loads
Sflexed ¼ 14:908 þ 23:810  T þ 12:077  M  146:166  LP were used to verify whether or not the recommended weight limit
(RWL) estimated by the NIOSH Lifting Equation (NLE) for a wide
þ 3:385  DS  0:146  T2 þ 0:091  M2 þ 49:924
range of lifting activities would result in L5-S1 compressive loads in
 ðLPÞ2 þ 0:013  Ds 2 þ 0:195  TM  2:113  TðLPÞ agreement with its recommended action limit of 3400 N. Moreover,
the L5-S1 shear forces under these lifting activities were also esti-
 0:015  TDS  3:034  MðLPÞ þ 0:936  MDS
mated by the predictive equations and compared to the threshold
 0:273  ðLPÞDS of 1000 N recommended in some studies in the literature. Results
(4) indicated that while the RWL estimated by the NLE was conser-
vative (L5-S1 compressive loads < 2000 N) for lifting tasks per-
where Cupright, Supright, Cflexed, Sflexed are the L5-S1 compressive and formed in the upright standing posture and those involving small
shear loads (N) in upright and flexed postures, respectively. T, M, trunk flexion angles (T < 30 where the L5-S1 compressive loads
and DS are in degree, kg, and cm, respectively and the spine loads were generally < 3400 N) it was too large for tasks involving
are calculated in N. The lumbopelvic ratio is calculated as the moderate to large trunk flexion (T  30 ). Furthermore, while the
lumbar rotation divided by the pelvis rotation where the former is RWL estimated by the NLE was adequate (L5-S1 shear
equal to the algebraic difference between measured trunk (T) and loads < 1000 N) for lifting tasks performed in the upright standing
pelvis (P) rotations for the lifting activity under consideration posture it was not so in flexed postures. Therefore, the NLE may not
 N. Arjmand et al. / International Journal of Industrial Ergonomics 47 (2015) 1e8 5

Table 2 In this study, the spinal loads were estimated through the pre-
Recommended weight limit (RWL) of the NIOSH Lifting Equations for lifting activ- dictive equations that are developed based on a complex detailed
ities in flexed postures compared to the maximum weight in hands (M3400) that
results in the NIOSH L5-S1 compression force limit of 3400 N.
trunk finite element biomechanical model while considering a
constant body weight (68.4 kg) and height (174.5 cm). This body
Task V H DL5-S1 T P RWL M3400 RWL/M3400 weight is below that of the 50th percentile of American females
(cm) (cm) (cm) (deg) (deg) (kg) (kg)
(McDowell et al., 2008). Our recent detailed modeling studies
7 79.5 25 35.6 10 2.6 22.7 24.8 0.9 (Arjmand et al., 2014a; Hajihosseinali et al., 2015) as well as those of
8 30 40.6 18.9 22.1 0.9
Han et al. (2013) have indicated the crucial role of body weight on
9 40 50.6 14.2 17.9 0.8
10 50 60.6 11.3 14.8 0.8 predicted spinal loads. As body weight almost doubles from 56 to
11 77.9 25 37.7 20 6.9 22.8 21.6 1.1 110 kg, our simulations while properly scaling musculature anat-
12 30 42.7 19.0 19.2 1.0 omy based on in vivo data (Anderson et al., 2011) yield increases of
13 40 52.7 14.2 15.4 0.9
20e120% in spinal loads depending on the lifting activity (Arjmand
14 50 62.7 11.4 12.6 0.9
15 75.3 25 39.8 30 9.5 23.0 19.3 1.2
et al., 2014a; Hajihosseinali et al., 2015). Therefore, had a body
16 30 44.8 19.1 17.0 1.1 weight of 84 kg and the same height of ~175 cm (corresponding to
17 40 54.8 14.4 13.6 1.1 50th percentile of North American males (McDowell et al., 2008))
18 50 64.8 11.5 11.1 1.0 for instance been considered, substantially larger spinal loads
19 71.4 25 42.2 40 12.4 22.8 16.8 1.4
compared to those reported (Table 1) would have been predicted.
20 30 47.2 19.0 14.8 1.3
21 40 57.2 14.2 11.8 1.2 The inadequacy of the NLE in controlling spine compressive load at
22 50 67.2 11.4 9.5 1.2 3400 N would in this case extend to additional lifting tasks
23 66.1 25 45.0 50 14.4 22.4 14.5 1.5 considered in this study.
24 30 50.0 18.7 12.8 1.5 On the other hand, had a smaller body weight of ~56 kg (cor-
25 40 60.0 14.0 10.2 1.4
26 50 70.0 11.2 8.2 1.4
responding to 25th percentile of North American females (Waters
27 60.6 25 47.1 60 16.8 22.0 12.8 1.7 et al., 1993)) for instance been considered, lower (by 15% at most)
28 30 52.1 18.3 11.3 1.6 L5-S1 compressive loads (Arjmand et al., 2014a; Hajihosseinali et
29 40 62.1 13.8 9.0 1.5 al., 2015) compared to those listed in Table 1 would have been
30 50 72.1 11.0 7.3 1.5
predicted. In this case, the L5-S1 compression in all tasks involving
31 53.7 25 48.5 70 22.2 21.5 11.4 1.9
32 30 53.5 17.9 10.1 1.8 trunk flexion angle of 50 and larger (except in task 26) would
33 40 63.5 13.5 8.0 1.7 exceed 3400 N. Finally, it is to be noted that the NLE is applied to all
34 50 73.5 10.8 6.5 1.7 workers equally regardless of their individual body weight.
35 47.1 25 49.9 80 28.3 21.1 10.4 2.0 Therefore, based on the representative tasks considered in this
36 30 54.9 17.6 9.3 1.9
37 40 64.9 13.2 7.4 1.8
study, our conclusion that the RWL in the NLE may generate spine
38 50 74.9 10.5 6.0 1.8 loads beyond recommended limits remains valid irrespective of the
39 40.7 25 51.2 90 35 20.6 9.9 2.1 body weight considered. Also, in our in vivo study the subject was
40 30 56.2 17.2 8.8 2.0 instructed to stretch his arms in the gravity direction while V
41 40 66.2 12.9 7.1 1.8
(vertical position of the hand load with respect to the inter-feet
42 50 76.2 10.3 5.9 1.7
43 32.9 25 50.7 100 41 20.1 10.2 2.0 midpoint) was measured for different trunk rotations (T ¼ 10, 20,
44 30 55.7 16.7 9.2 1.8 30, …, 100, and 107.6 ). It is important to note that the measured V
45 40 65.7 12.6 7.5 1.7 and the corresponding T represent one plausible lifting posture
46 50 75.7 10.0 6.2 1.6 among many others as the arms flex or knees bend.
47 27.7 25 49.4 107.6 46 19.7 10.9 1.8
48 30 54.4 16.4 9.8 1.7
The tasks investigated in the present study represent some
49 40 64.4 12.3 8.1 1.5 typical ones performed in an occupational environment. It is
50 50 74.4 9.9 6.8 1.5 important to note that for a given task (i.e., for an identical trunk
flexion angle, hand load magnitude, and horizontal load distance to
the L5-S1) individuals may adapt alternative postures with
adequately control spinal loads as intended in protecting the spinal different lumbopelvis (LP) ratios (or different pelvis rotations than
column during lifting activities in flexed postures. those given in Table 1). As the LP ratio is an input into our predictive
equations in flexed postures, its variation can affect the predicted
spinal loads. Our sensitivity analyses on the effect of variations in LP
4.1. Methodological issues ratio on the predicted L5-S1 compressive force, however, indicate
that the predicted spinal loads are only slightly altered with the LP
Both our in vivo and biomechanical modeling (predictive ratio. For instance, in tasks (47e50) in which the pelvic rotation is
equations) studies have some limitations. In our in vivo study to taken as 46 (Table 1), a substantial reduction by 20 (from 56 or
measure trunk kinematics, for a given trunk flexion angle, the 36 ) equivalent to a marked 115% increase in LP ratio (from 0.92 to
pelvis rotations needed for the estimation of spine loads in pre- 1.99) decreases the L5-S1 compression only by 5% at most.
dictive equations were not measured with loads in hands equal to The predictive equations are developed based on a detailed
the calculated RWL. It was therefore assumed that for a given lifting musculoskeletal model of the spine whose predictions for the L4-L5
activity under a specific flexed trunk angle (T), the pelvis rotation intradiscal pressure and back muscle forces for a wide range of
(P) and thus the LP ratio would not have altered had the RWL been lifting activities are found in agreement with in vivo measured
handled in these lifting tasks. It has been shown that holding 10 kg intradiscal pressure values and qualitatively with electrographic
in hands can increase the LP ratio during forward bending activities (EMG) data (Arjmand and Shirazi-Adl, 2006; Arjmand et al., 2007,
by about 15% at most (Granata and Sanford, 2000). A sensitivity 2009, 2010; 2011). This model similar to its counterparts, however,
analysis on the effect of LP ratio on spinal loads estimated by our has its own assumptions that are discussed in details elsewhere
predictive equations showed that increasing the LP ratio at any (Arjmand and Shirazi-Adl, 2006). For example, abdominal antago-
trunk angle by 15% influences the L5-S1 compressive and shear nistic coactivities are neglected in the model and thus in predictive
loads by less than 1 and 5%, respectively. equations. Also, the model is not sensitive to the vertical position of
6 N. Arjmand et al. / International Journal of Industrial Ergonomics 47 (2015) 1e8

weights in hands in upright postures, i.e., the same spinal loads are committee diminished the RWL by 22.5% versus lifts at the waist
predicted by the model as load height increases (tasks 3 through 6 level (V ¼ 75 cm) ((i.e., VM ¼ 1(0.003  jV75j)). Revision of the
in Table 1) at a constant horizontal distance to the body. This is in NLE by the introduction of a trunk flexion angle multiplier can
disagreement with in vivo studies (Arjmand et al., 2009; Granata hence markedly improve the estimations of RWL.
and Orishimo, 2001) indicating that back and abdominal muscle The NLE does not adequately control the L5-S1 compressive load
EMG activities increase as load height increases (with a constant in lifting activities involving moderate to large trunk flexion as it
lever arm) which is due to the stability requirements. The predicted underestimates the importance of trunk flexion angle on the spine
spine loads during these lifting activities with the corresponding compressive load. In agreement with our biomechanical modeling
RWL in hands are however well below the recommended permis- simulations (Arjmand and Shirazi-Adl, 2006; Arjmand et al., 2010),
sible limits (Table 1). in vivo studies also point to the great impact of trunk flexion angle
on spine loads by measuring ~230% increase (from 1 to 2.3 MPa) in
4.2. Analysis of results the L4-L5 intradiscal pressure (as an indicator of the spine
compressive load) as trunk flexes forward from upright to ~70
Analysis of results indicated that the vertical multiplier (VM) in flexion with 20 kg in hands or by ~255% (from 0.45 to 1.6 MPa) as
NLE was a primary reason for the inadequate estimation of RWL in trunk flexes forward from upright to full flexion without load in
lifting tasks with flexed postures while the horizontal multiplier hands (Wilke et al., 2001). This illustrates the importance of
(HM) properly modulated RWL. According to our predictive equa- increasing the height of the lift (or decreasing trunk flexion) in an
tions during lifting activities in flexed postures as horizontal dis- ergonomic intervention in order to reduce external back loading as
tance of weights in hands increased from H ¼ 25 cm to H ¼ 50 cm demonstrated in a recent manual material handling study
the maximum hand load (M3400) should have decreased by ~40% in (Plamondon et al., 2012). This can also be achieved by adapting a
order to maintain the L5-S1 compressive load at 3400 N (Fig. 2). The squat lifting technique that involves relatively smaller trunk angles
horizontal multiplier (HM ¼ 25/H) also reduced from 1 to 0.5 as when compared to a stoop one. Our previous in vivo-modeling
load moved away from H ¼ 25e50 cm thus decreasing the weight investigations have indicated that spinal compression/shear forces
limit by about 50%. The NIOSH horizontal multiplier, hence, are smaller in squat lifts than in stoop ones (Bazrgari et al., 2008).
appropriately controlled the L5-S1 compressive load to meet the The inadequacy in the vertical NLE could partly be due to the
recommended limit. fact that the 1991 NIOSH committee used very simplistic biome-
On the other hand, our predictive equations showed that at any chanical models with limited number of trunk muscles (e.g., one
given horizontal load distance, H, as the vertical distance of the equivalent muscle for synergetic muscles with linear line of action
weight in hands reduced from V ¼ 79.5 cm (at T ¼ 10 ) to and equilibrium at a single level) for the estimation of the L5-S1
V ¼ 47.1 cm (at T ¼ 80 ) the maximum weight (M3400) should have compressive loads (Waters et al., 1993). The predictive equations,
decreased by ~60% in order to maintain the L5-S1 compressive load however, are developed based on a detailed multi-level biome-
at 3400 N (Fig. 3). The NIOSH vertical multiplier chanical model in which much more accurate musculature ar-
(VM ¼ 1(0.003  jV75j)), however, slightly reduced from ~0.987 chitecture (52 trunk muscle fascicles each with its own line of
to ~0.916 (Table 1) thereby decreasing the weight by only ~7% as action), their curved-shape (wrapping) geometry during flexion
load vertical distance dropped from 79.5 to 47.1 cm (or equivalently postures, and nonlinear passive properties of the ligamentous
trunk flexion angle increased from 10 to 80 ). The difference be- spine are incorporated (Arjmand and Shirazi-Adl, 2006; Arjmand
tween RWL and M3400 increased (i.e., inadequacy of the NLE in et al., 2006, 2011). This is the most detailed biomechanical model
controlling spine loads grew) as the trunk flexion angle increased; ever used to investigate whether or not the NLE maintains spinal
reaching its peak at the trunk flexion angle of 90 (e.g., RWL/ loads below the recommended limits. It is evident that a simplistic
M3400 ¼ 2.1) (Table 2). The 1991 NIOSH committee has acknowl- model such as the L5-S1 model of 3DSSPP software (University of
edged that for lifting activities near the floor the biomechanical Michigan Center for Ergonomics, 2014) likely yields spinal loads
studies suggest an increased lumbar load and physiological studies different from those listed in Table 1. In a recent study, Potvin
indicate an increased energy expenditure (Waters et al., 1993). (2014) states that RWLs from NLE yield L5-S1 compression
However, due to the lack of direct empirical data to provide a forces close to the limit of 3400 N for lifts below knuckle and
specific reduction in RWL for lifts at floor level (V ¼ 0), the much lower compression forces for lifts at higher height. These
findings are, however, based on a very simplistic biomechanical
model of the spine assuming a constant moment arm of 6 cm for a

Fig. 2. Variation of the maximum handled weight (M3400) that preserves the L5-S1 Fig. 3. Variation of the maximum handled weight (M3400) that preserves the L5-S1
compressive load at 3400 N with the horizontal position of the weight (H) at different compressive load at 3400 N with the vertical position of the weight (V) at different
vertical positions of load (V). horizontal positions of weight (H).
 N. Arjmand et al. / International Journal of Industrial Ergonomics 47 (2015) 1e8 7

single extensor muscle incorporated in the model. In a recent dependence on the individual's body weight, body height, gender,
study (Arjmand et al., 2014b; Rajaee et al., 2015), we have shown and age (Jager and Luttmann, 1991; Genaidy et al., 1993) . Ergono-
that different lifting analysis tools such as AnyBody Modeling mist and occupational health practitioners are advised to use the
System (Damsgaard et al., 2006), 3DSSPP (University of Michigan NLE along with the predictive equations of the spine loads pro-
Center for Ergonomics, 2014), our predictive equations, HCBCF posed in this study when evaluating the risk of injury to the spine in
equation (Merryweather et al., 2009), and simple polynomial lifting activities particularly those involving moderate to large
model (McGill et al., 1996) could yield very different spinal loads trunk flexion. The NLE may be used for lifting activities in upright
and hence risk of injury when applied to an identical task. This is postures as well as lifting tasks involving small trunk flexion such
due to distinct underlying assumptions and simplifications used as those performed with a squat technique.
in these models. Among the forgoing five tools, the Anybody
Modeling System and our predictive equations estimated intra- Acknowledgment
discal pressures in closer agreement with available in vivo mea-
surements (RMSE z 0.12 MPa). This work is supported by grants from Sharif University of
Our biomechanical modeling analysis demonstrated that the L5- Technology (Tehran, Iran) and Institut de recherche Robert-Sauve

S1 intervertebral disc is the site of greatest lumbar posterior-
et en se
en sante
curite

du travail, IRSST (grant number: 2010-0023)
anterior shear force due to its initial forward slope in the sagittal (Montre
al, Canada).
plane (Arjmand et al., 2006). The L5-S1 shear load predicted based
on the RWL exceeded 1000 N suggested as a limit in some earlier References
studies (Gallagher and Marras, 2012; McGill, 1997; McGill et al.,
1998) but remained below the in vitro measured yield shear National Research Council & Institute of Medicine [NRC-IOM], 2001. Musculoskel-
forces of 1.57 ± 0.11 kN (Skrzypiec et al., 2012) in all lifting activities etal Disorders and the Workplace: Low Back and Upper Extremities. National
Academy Press, Washington, DC.
involving trunk flexion (Table 1). In vitro ultimate shear forces of up Anderson, D.E., D'Agostino, J.M., Bruno, A.G., Manoharan, R.K., Bouxsein, M.L., 2011.
to 2.8 kN (Cyron et al., 1976), 3.29 ± 0.16 kN (Skrzypiec et al., 2012), Regressions for estimating muscle parameters in the thoracic and lumbar trunk
and 2.24 ± 0.57 kN (Frei et al., 2002) are reported using different for use in musculoskeletal modeling. J. Biomech. 45 (1), 66e75.
Arjmand, N., Shirazi-Adl, A., 2006. Model and in vivo studies on human trunk load
experimental setups. A recent review study suggests the average partitioning and stability in isometric forward flexions. J. Biomech. 39 (3),
ultimate shear forces of 1.9 and 1.7 kN for working age males and 510e521.
females, respectively (Gallagher and Marras, 2012) and recom- Arjmand, N., Shirazi-Adl, A., Bazrgari, B., 2006. Wrapping of trunk thoracic extensor
muscles influences muscle forces and spinal loads in lifting tasks. Clin. Biomech.
mends the 1000 N limit for occasional exposure to shear which is
21, 668e675.
also in agreement with the National Research Council & Institute of Arjmand, N., Shirazi-Adl, A., Parnianpour, M., 2007. Trunk biomechanical models
Medicine (NRC/IOM) Report (NRC-IOM 2001) as well as the limit based on equilibrium at a single-level violate equilibrium at other levels. Eur.
reported elsewhere (McGill, 1997; McGill et al., 1998). It is, however, Spine J. 16 (5), 701e709.
Arjmand, N., Gagnon, D., Plamondon, A., Shirazi-Adl, A., Larivie re, C., 2009. Com-
to be noted that the L5-S1 shear yield and strength forces are likely parison of trunk muscle forces and spinal loads estimated by two biome-
different from those at the upper lumbar levels due to substantial chanical models. Clin. Biomech. 24 (7), 533e541.
Arjmand, N., Gagnon, D., Plamondon, A., Shirazi-Adl, A., Larivie re, C., 2010.
changes in disc size, facet orientations and compression preloads. It
A comparative study of two trunk biomechanical models under symmetric and
is to be noted that, apart from the segmental level, the spinal asymmetric loadings. J. Biomech. 43, 485e491.
strength in shear likely depends also on the compression preload Arjmand, N., Plamondon, A., Shirazi-Adl, A., Larivie re, C., Parnianpour, M., 2011.
and posture (Howarth and Callaghan, 2013). Predictive equations to estimate spinal loads in symmetric lifting tasks.
J. Biomech. 44, 84e91.
Some epidemiological studies suggest a relationship between Arjmand, N., Hajihosseinali, M., Shirazi-Adl, A., 2014aa. Effect of body weight on
the LI and the prevalence of LBP (statistically significant increase in spine loads in different activities: a detailed biomechanical modeling investi-
prevalence as LI exceeds 1.0 or 2.0) (e.g., Waters et al., 2011; Lu et al., gation. In: The 7th World Congress of Biomechanics, Session 20-14, Boston (July
2014).
2014). On the other hand, the NLE application to predict the inci-
Arjmand, N., Rajaee, M.A., Shirazi-Adl, A., Plamondon, A., 2014bb. Comparison of
dence of low back disorders has been questioned; no relations have five lifting analysis tools to determine low back loads. In: The 7th World
been detected between the LI and either the incidence of lost time Congress of Biomechanics, Session 21-14, Boston (July 2014).
Bazrgari, B., Shirazi-Adl, A., Arjmand, N., 2008. Analysis of squat and stoop dynamic
due to LBP or the prevalence of LBP (Pinder and Frost, 2011). Our
liftings: muscle forces and internal spinal loads. Eur. Spine J. 16, 687e699.
findings here suggest that a modification in the NIOSH vertical Cyron, B.M., Hutton, W.C., Troup, J.D., 1976. Spondylolytic fractures. J. Bone Jt. Sur-
multiplier for tasks performed with moderate to large trunk angles gury Br. 58-B, 462e466.
is necessary in order to maintain the L5-S1 compression below Damsgaard, M., Rasmussen, J., Christensen, S.T., Surma, E., de Zee, M., 2006. Analysis
of musculoskeletal systems in the anybody modeling System. Simul. Model.
3400 N. In parallel and in light of data on spinal strength in Pract. Theory 14, 1100e1111.
compression and its dependence on the worker's body height, Frei, H., Oxland, T.R., Nolte, L.P., 2002. Thoracolumbar spine mechanics contrasted
weight, gender, and sex (Jager and Luttmann, 1991; Genaidy et al., under compression and shear loading. J. Orthop. Res. 20, 1333e1338.
Gallagher, S., Marras, W.S., 2012. Tolerance of the lumbar spine to shear: a review
1993), the limit of 3400 N set on the compression force should and recommended exposure limits. Clin. Biomech. 27 (10), 973e978.
also be re-visited The biomechanics and epidemiologic bases of this Garg, A., Moore, J.S., 1992. Epidemiology of low-back pain in industry. Occup. Med. 7
limit has indeed been questioned (Jager and Luttmann, 1999). (4), 593e608.
Genaidy, A.M., Waly, S.M., Khalil, T.M., Hidalgo, J., 1993. Spinal compression toler-
ance limits for the design of manual material handling operations in the
4.3. Conclusion workplace. Ergonomics 36, 415e434.
Granata, K.P., Orishimo, K.F., 2001. Response of trunk muscle coactivation to
changes in spinal stability. J. Biomech. 34, 1117e1123.
The NLE failed to adequately control spine compressive and Granata, K.P., Sanford, A.H., 2000. Lumbar-pelvic coordination is influenced by
shear loads in lifting activities involving moderate to large trunk lifting task parameters. Spine 25 (11), 1413e1418.
forward flexion. The NIOSH vertical multiplier was in part found Hajihosseinali, M., Arjmand, N., Shirazi-Adl, A., 2015. Effect of body weight on spinal
loads in various activities: A personalized biomechanical modeling approach.
responsible for this inadequacy while the horizontal multiplier
J. Biomech. 48, 276e282.
correctly modulated the hand load. In order to address this short- Han, K.S., Rohlmann, A., Zander, T., Taylor, W.R., 2013. Lumbar spinal loads vary with
coming in the NLE, it is suggested that a new multiplier accounting body height and weight. Med. Eng. Phys. 35 (7), 969e977.
for the trunk flexion angle be considered. In addition, the recom- Hoogendoorn, W.E., Bongers, P.M., de Vet, H.C., Douwes, M., Koes, B.W.,
Miedema, M.C., Arie €ns, G.A., Bouter, L.M., 2000. Flexion and rotation of the
mended compression limit of 3400 N can be revised in accordance trunk and lifting at work are risk factors for low back pain: results of a pro-
with the data on the spine compression strength and its spective cohort study. Spine 25 (23), 3087e3092.
8 N. Arjmand et al. / International Journal of Industrial Ergonomics 47 (2015) 1e8

Howarth, S.J., Callaghan, J.P., 2013. Compressive force magnitude and intervertebral reporting of low back pain in the automotive industry. Clin. Biomech. 13 (8),
joint flexion/extension angle influence shear failure force magnitude in the 561e573.
porcine cervical spine. J. Biomech. 45 (3), 484e490. Pascual, S.A., Naqvi, S., 2008. An investigation of ergonomics analysis tools used in
Jager, M., Luttmann, A., 1991. Compressive strength of lumbar spine elements industry in the identification of work-related musculoskeletal disorders. Int. J.
related to age, gender, and other influencing factors. In: Anderson, P.A., Occup. Saf. Ergon. 14 (2), 237e245.
hobart, ill, Danoff, J.V. (Eds.), Electromyo-graphical Kinesiology. Elsevier Sci- Pinder, A., Frost, G., 2011. Prospective Evaluation of the 1991 NIOSH Lifting Equation.
ences, Amsterdam, pp. 291e294. HSE Books (RR901). http://www.hse.gov.uk/research/rrhtm/rr901.htm.
Jager, M., Luttmann, A., 1999. Critical survey on the biomechanical criterion in the Plamondon, A., Lariviere, C., Delisle, A., Denis, D., Gagnon, D., 2012. Relative
NIOSH method for the design and evaluation of manual lifting tasks. Int. J. Ind. importance of expertise, lifting height and weight lifted on posture and lumbar
Ergon. 23, 331e337. external loading during a transfer task in manual material handling. Ergo-
Lu, M.L., Waters, T.R., Krieg, E., Werre, D., 2014. Efficacy of the revised NIOSH lifting nomics 55, 87e102.
equation to predict risk of low-back pain associated with manual lifting: a one- Potvin, J., 2014. Comparing the revised NIOSH lifting equation to the psychophys-
year prospective study. Hum. Factors 56, 73e85. ical, biomechanical and physiological criteria used in its development. Int. J. Ind.
Manchikanti, L., 2000. Epidemiology of low back pain. Pain Physician 3 (2), Eng. 44 (2), 246e252.
167e192. Rajaee, M.A., Arjmand, N., Shirazi-Adl, A., Plamondon, A., Schmidt, H., 2015.
McDowell, M.A., Fryar, C.D., Ogde, C.L., Flegal, K.M., 2008. Anthropometric Comparative evaluation of six quantitative lifting tools to estimate spine loads
Reference Data for Children and Adults: United States, 2003e2006. National during static activities. Appl. Ergon. 48, 22e32.
Center for Health Statistics, Hyattsville, MD. National health statistics reports; Skrzypiec, D.M., Klein, A., Bishop, N.E., Stahmer, F., Püschel, K., Seidel, H.,
no10. Morlock, M.M., Huber, G., 2012. Shear strength of the human lumbar spine. Clin.
McGill, S.M., 1997. The biomechanics of low back injury: implications on current Biomech. 27 (7), 646e651.
practice in industry and the clinic. J. Biomech. 30 (5), 465e475. Tafazzol, A., Arjmand, N., Shirazi-Adl, A., Parnianpour, M., 2014. Lumbopelvic
McGill, S.M., Norman, R.W., Cholewicki, J., 1996. A simple polynomial that predicts rhythm during forward and backward sagittal trunk rotations: combined
low-back compression during complex 3-D tasks. Ergonomics 39, 1107e1118. in vivo measurement with inertial tracking device and biomechanical
McGill, S.M., Norman, R.W., Yingling, V.R., Wells, R.W., Neumann, P., 1998. “Shear modeling. Clin. Biomech. 29 (1), 7e13.
happens! Suggested guidelines for ergonomists to reduce the risk of low back University of Michigan Center for Ergonomics, 2014. 3D Static Strength Prediction
injury from shear loading. In: Proceedings of the 30th Annual Conference of the Program User's Manual.
Human Factors Association of Canada (HFAC), Mississauga, Ontario, Canada. Waters, T.R., Putz-Anderson, V., Garg, A., Fine, L.J., 1993. Revised NIOSH equation for
Merryweather, A.S., Loertscher, M.C., Bloswick, D.S., 2009. A revised back the design and evaluation of manual lifting tasks. Ergonomics 36 (7), 749e776.
compressive force estimation model for ergonomic evaluation of lifting tasks. Waters, T.R., Baron, S.L., Kemmlert, K., 1998. Accuracy of measurements for the
Work 34, 263e272. revised NIOSH lifting equation. National Institute for occupational safety and
Nachemson, A., 1981. Disc pressure measurements. Spine 6 (1), 93e97. Health. Appl. Ergon. 29 (6), 433e438.
Van Nieuwenhuyse, Fatkhutdinova, A.L., Verbeke, G., Pirenne, D., Johannik, K., Waters, T.R., Lu, M.L., Piacitelli, L.A., Werren, D., Deddens, J.A., 2011. Efficacy of the
Somville, P.R., Mairiaux, P., Moens, G.F., Masschelein, R., 2004. Risk factors for revised NIOSH lifting equation to predict risk of low back pain due to manual
first-ever low back pain among workers in their first employment. Occup. Med. lifting: expanded cross-sectional analysis. J. Occup. Environ. Med. 9, 1061e1067.
54 (8), 513e519. Wilke, H., Neef, P., Hinz, B., Seidel, H., Claes, L., 2001. Intradiscal pressure together
Norman, R., Wells, R., Neumann, P., Frank, J., Shannon, H., Kerr, M., 1998. with anthropometric datada data set for the validation of models. Clin. Bio-
A comparison of peak vs cumulative physical work exposure risk factors for the mech. 16, S111eS126.